:>"^Z0ã@súdZddlmZddlmZddlmZdd„ZGdd„dejƒZGd d „d ejƒZGd d „d e ƒZ d d„Z dd„Z dd„Z dd„Zdd„Zdd„Zdd„Zdd„Zdd„ZdS) z,Module for the support of MEME motif format.é)Úprint_function)ÚSeq)Úmotifsc Cs˜tƒ}t||ƒt||ƒt||ƒt||ƒt||ƒx*|D]}|jdƒrQPqQWtdƒ‚|j}d|j k}xt |ƒ\}}}}t |ƒ} t || |||ƒ} t || ƒ} || _|| _|| _| | _|j| ƒt|ƒ|ks&t‚t|ƒyt|ƒ}Wntk r`tdƒ‚YnX|jdƒrqP|jdƒs’td|ƒ‚q’W|S)a¨Parse the text output of the MEME program into a meme.Record object. Examples -------- >>> from Bio.motifs import meme >>> with open("motifs/meme.out") as f: ... record = meme.read(f) >>> for motif in record: ... for instance in motif.instances: ... print(instance.motif_name, instance.sequence_name, instance.strand, instance.pvalue) Motif 1 SEQ10; + 8.71e-07 Motif 1 SEQ9; + 8.71e-07 Motif 1 SEQ8; + 8.71e-07 Motif 1 SEQ7; + 8.71e-07 Motif 1 SEQ6; + 8.71e-07 Motif 1 SEQ5; + 8.71e-07 Motif 1 SEQ4; + 8.71e-07 Motif 1 SEQ3; + 8.71e-07 Motif 1 SEQ2; + 8.71e-07 Motif 1 SEQ1; + 8.71e-07 zMOTIF zUnexpected end of streamÚrevcompzNUnexpected end of stream: Expected to find new motif, or the summary of motifszSUMMARY OF MOTIFSÚMOTIFz$Line does not start with 'MOTIF': %s)ÚRecordÚ__read_versionÚ__read_datafileÚ__read_alphabetÚ__read_sequencesÚ__read_commandÚ startswithÚ ValueErrorÚalphabetÚcommandÚ__read_motif_statisticsÚ__read_motif_nameÚ__read_motif_sequencesÚMotifÚlengthÚnum_occurrencesÚevalueÚnameÚappendÚlenÚAssertionErrorÚ__skip_unused_linesÚnextÚ StopIteration) ÚhandleÚrecordÚlinerrÚ motif_numberrrrrÚ instancesÚmotif©r%ú4/tmp/pip-build-ww9dw3qa/biopython/Bio/motifs/meme.pyÚreadsB                 r'c@s(eZdZdZdddd„ZdS)rzäA subclass of Motif used in parsing MEME (and MAST) output. This subclass defines functions and data specific to MEME motifs. This includes the motif name, the evalue for a motif, and its number of occurrences. NcCsGtjj|||ƒd|_d|_d|_d|_d|_dS)zInitialize the class.grN)rrÚ__init__rrrÚidZalt_id)Úselfrr#r%r%r&r(Qs     zMotif.__init__)Ú__name__Ú __module__Ú __qualname__Ú__doc__r(r%r%r%r&rIs rc@s"eZdZdZdd„ZdS)ÚInstancezGA class describing the instances of a MEME motif, and the data thereof.cOsPtjj|||Žd|_d|_d|_d|_d|_d|_dS)zInitialize the class.Úrgð?N)rr(Ú sequence_nameÚstartÚpvalueÚstrandrÚ motif_name)r*ÚargsÚkwdsr%r%r&r(^s     zInstance.__init__N)r+r,r-r.r(r%r%r%r&r/[s r/c@s.eZdZdZdd„Zdd„ZdS)raPA class for holding the results of a MEME run. A meme.Record is an object that holds the results from running MEME. It implements no methods of its own. The meme.Record class inherits from list, so you can access individual motifs in the record by their index. Alternatively, you can find a motif by its name: >>> from Bio import motifs >>> with open("motifs/meme.out") as f: ... record = motifs.parse(f, 'MEME') >>> motif = record[0] >>> print(motif.name) Motif 1 >>> motif = record['Motif 1'] >>> print(motif.name) Motif 1 cCs1d|_d|_d|_d|_g|_dS)z Initialize.r0N)ÚversionÚdatafilerrÚ sequences)r*r%r%r&r(~s     zRecord.__init__cCsJt|tƒr6x4|D]}|j|kr|SqWntj||ƒSdS)zReturn the motif of index key.N)Ú isinstanceÚstrrÚlistÚ __getitem__)r*Úkeyr$r%r%r&r>†s   zRecord.__getitem__N)r+r,r-r.r(r>r%r%r%r&ris  rcCsVx*|D]}|jdƒrPqWtdƒ‚|jƒ}|jƒ}|d|_dS)Nz MEME versionzFImproper input file. File should contain a line starting MEME version.é)r rÚstripÚsplitr8)r rr!Úlsr%r%r&r“s    rcCsøx*|D]}|jdƒrPqWtdƒ‚yt|ƒ}Wntk r]tdƒ‚YnX|jdƒs}td|ƒ‚yt|ƒ}Wntk r­tdƒ‚YnX|jdƒsÍtd|ƒ‚|jƒ}|jd d ƒ}||_dS) Nz TRAINING SETz‰Unexpected end of stream: 'TRAINING SET' not found. This can happen with minimal MEME files (MEME databases) which are not supported yet.zDUnexpected end of stream: Expected to find line starting with '****'z****z#Line does not start with '****': %szHUnexpected end of stream: Expected to find line starting with 'DATAFILE'ZDATAFILEz'Line does not start with 'DATAFILE': %sz DATAFILE= r0)r rrrrAÚreplacer9)r rr!r%r%r&r žs(     r c Cs«yt|ƒ}Wntk r0tdƒ‚YnX|jdƒsPtd|ƒ‚|jƒ}|jddƒ}|dkrƒd}n|dkr˜d}nd}||_dS) NzHUnexpected end of stream: Expected to find line starting with 'ALPHABET'ZALPHABETz'Line does not start with 'ALPHABET': %sz ALPHABET= r0ZACGTZACGUZACDEFGHIKLMNPQRSTVWY)rrrr rArDr)r rr!Úalr%r%r&r ·s      r cCs#yt|ƒ}Wntk r0tdƒ‚YnX|jdƒsPtd|ƒ‚yt|ƒ}Wntk r€tdƒ‚YnX|jdƒs td|ƒ‚x||D]h}|jdƒr½P|jƒ}|jƒ}|jj|dƒt|ƒd kr§|jj|d ƒq§Wtd ƒ‚dS) NzMUnexpected end of stream: Expected to find line starting with 'Sequence name'z Sequence namez,Line does not start with 'Sequence name': %szDUnexpected end of stream: Expected to find line starting with '----'z----z#Line does not start with '----': %sz***réézCUnexpected end of stream: Expected to find line starting with '***') rrrr rArBr:rr)r rr!rCr%r%r&r És*     r cCsXx*|D]}|jdƒrPqWtdƒ‚|jƒ}|jddƒ}||_dS)Nzcommand:zGUnexpected end of stream: Expected to find line starting with 'command'z command: r0)r rrArDr)r rr!r%r%r&r âs   r c Cs\|jƒ}|ddks"t‚|ddd…dkr\t|djdƒdƒ}nt|dƒ}|djdƒr’|d d…}n|dd…}|ddd …}|ddd …}|ddd …}|d d d d gksùt‚x |D]}|dkst‚qWt|dƒ}t|dƒ} t|d ƒ} ||| | fS)Nrrr@ézMEME-ú-éZMEMErGÚwidthÚsitesZllrzE-valueú=)rBrÚintr Úfloat) r!Úwordsr"Z key_valuesÚkeysZ equal_signsÚvaluesZ equal_signrrrr%r%r&rís$   rcCs_x'|D]}d|krPqWtdƒ‚|jƒ}|jƒ}dj|dd…ƒ}|S)Nzsorted by position p-valuez3Unexpected end of stream: Failed to find motif nameú rr@)rrArBÚjoin)rr!rPrr%r%r&r s     rc Csõyt|ƒ}Wntk r0tdƒ‚YnX|jdƒsPtd|ƒ‚yt|ƒ}Wntk r€tdƒ‚YnX|jdƒs td|ƒ‚yt|ƒ}Wntk rÐtdƒ‚YnX|jdƒsðtd|ƒ‚g}xì|D]Ø}|jdƒrP|jƒ}|jƒ}|rC|jdƒ}nd}|d } t| ƒ|kskt‚t | |ƒ} || _ |d | _ t |dƒ| _ t|d ƒ| _|| _|| _|j| ƒqýWtd ƒ‚tj||ƒS) Nz8Unexpected end of stream: Failed to find motif sequencesz---z"Line does not start with '---': %szMUnexpected end of stream: Expected to find line starting with 'Sequence name'z Sequence namez,Line does not start with 'Sequence name': %srJú+érr@zUnexpected end of stream)rrrr rArBÚpoprrr/r5r1rNr2rOr3r4rrrZ Instances) rr5rrrr!r#rPr4ÚsequenceÚinstancer%r%r&rsN            rcCsËx*|D]}|jdƒrPqWtdƒ‚x*|D]}|jdƒr4Pq4Wtdƒ‚x*|D]}|jdƒraPqaWtdƒ‚x*|D]}|jdƒrŽPqŽWtdƒ‚x*|D]}|jdƒr»Pq»Wtdƒ‚yt|ƒ}Wntk rtd ƒ‚YnX|jƒr.td |ƒ‚yt|ƒ}Wntk r^td ƒ‚YnX|jd ƒs~td |ƒ‚x'|D]}|jƒr…Pq…Wtd ƒ‚|jd ƒsÇtd |ƒ‚dS)Nzlog-odds matrixzOUnexpected end of stream: Expected to find line starting with 'log-odds matrix'z---zCUnexpected end of stream: Expected to find line starting with '---'zletter-probability matrixzYUnexpected end of stream: Expected to find line starting with 'letter-probability matrix'ZTimezDUnexpected end of stream: Expected to find line starting with 'Time'z5Unexpected end of stream: Expected to find blank linez Expected blank line, but got: %szCUnexpected end of stream: Expected to find line starting with '***'z***z"Line does not start with '***': %s)r rrrrA)rr!r%r%r&rBsL                rN)r.Ú __future__rZBiorrr'rr/r=rrr r r r rrrrr%r%r%r&Ús  ;*     ,